hyperp is a fully automated serverless compute platform built on AWS. It embraces a complete GitOps workflow, automatically managing CI/CD pipelines while providing integrated artifact storage for data transfer between jobs and local download. The platform includes built-in cost estimation for both compute and storage resources per run, and offers a straightforward CLI for easy interaction.

How it works

Setup instructions

• Monitor triggerred runs and view their details.
• View task logs.
• View cost estimations for compute resource usage per task/job/workflow.
• Get storage usage per job/workflow.
• Download artifacts to your local machine.

Hyperp consists of the following components:

• VPC: Large addressable VPC (10.0.0.0/16) with public subnets
• ECS Fargate Cluster: For running containerized jobs
• ECR Repository: For storing built Docker images
• EFS File System: For sharing artifacts between jobs
• DynamoDB: For storing workflow metadata, runs, and state
• S3 Bucket: For downloading artifacts
• EventBridge: For tracking ECS task state changes

1. GitHub Webhook Handler: Processes commit events, syncs workflows, triggers runs
2. Task State Change Handler: Tracks ECS task completion, manages job orchestration
3. EFS Controller: Creates directories on EFS for artifact storage
4. CLI REST API: Provides REST API endpoints for the CLI tool to query workflow runs